BACE is a very popular target (there are potentially 20 Billion reasons for it). As we noted in April, Pfizer has entered the contest (publicly now). Pfizer utilized a subset (340 fragments) of their recently described Fragment library (GFI) used X-ray as the discovery platform soaked in 4 at a time. At an intial concentration of 20 mM, 58 of the 85 mixes yielded useable diffraction data. They then repeated the screen at 2mM and 200uM to attempt to gather data on compounds which disrupted the crystal lattice at higher concentrations. All of this led to the discovery of one fragment (the spiropyrrolidine).
They then threw the biophysical and biochemical book at this compound to establish it as a bona fide inhibitor: Octet (1.4mM Kd, 0.31 kcal/mol/atom), STD-NMR, 1H-15N HSQC NMR, functional NMR assay to determine weak Kds, and a BACE inhibition assay (1.1mM). It was found that this fragment had excellent permeability and low potential PGP efflux.
They then did their SAR to optimize the core and develop "growth" vectors. They ended up improving potency by three orders of magnitude with seriously affecting the ligand efficiency nor the in vitro properties predictive of good brain penetration.
To me, this is the most interesting point in the paper. Does ontogeny recapitulate phylogeny for drugs? If you start with good properties, do you keep them? I know there is a good amount of debate on whether or not a fragment will keep its binding mode as it is optimized/expanded. What is the general thinking on properties? Do people screen (at least in silico) their libraries for things like permeability?